Description of problem:
One customer is seeing that some pods are hanging in a terminating state when he triggered a cluster upgrade or when he do a deployment rollout. The only way to get rid of this is to force terminate the pods. This can be observed for any pods irrespective of project and nodes.
Found a similar bug with 4.7 :
https://bugzilla.redhat.com/show_bug.cgi?id=1915085
Version-Release number of selected component (if applicable):
Current OCP version:
version 4.6.17
Additional info:
CU faced this issue with 4.5 initially then they upgraded to 4.6, but still they are facing this issue.
I will attach the logs.